The 'sysret' fastpath does not correctly restore even all regular registers, much less any segment registers or reflags values. That is very much part of why it's faster than 'iret'. Normally that isn't a problem, because the normal ptrace() interface catches the process using the signal handler infrastructure, which always returns with an iret. However, some paths can get caught using ptrace_event() instead of the signal path, and for those we need to make sure that we aren't going to return to user space using 'sysret'. Otherwise the modifications that may have been done to the register set by the tracer wouldn't necessarily take effect. Fix it by forcing IRET path by setting TIF_NOTIFY_RESUME from arch_ptrace_stop_needed() which is invoked from ptrace_stop().
Unified · Split
| @@ -231,6 +231,22 @@ static inline unsigned long regs_get_kernel_stack_nth(struct pt_regs *regs, · 230 unmodified lines | ||
| 231 | 231 | |
| 232 | 232 | #define ARCH_HAS_USER_SINGLE_STEP_INFO |
| 233 | 233 | |
| 234 | +/* | |
| 235 | + * When hitting ptrace_stop(), we cannot return using SYSRET because | |
| 236 | + * that does not restore the full CPU state, only a minimal set. The | |
| 237 | + * ptracer can change arbitrary register values, which is usually okay | |
| 238 | + * because the usual ptrace stops run off the signal delivery path which | |
| 239 | + * forces IRET; however, ptrace_event() stops happen in arbitrary places | |
| 240 | + * in the kernel and don't force IRET path. | |
| 241 | + * | |
| 242 | + * So force IRET path after a ptrace stop. | |
| 243 | + */ | |
| 244 | +#define arch_ptrace_stop_needed(code, info) \ | |
| 245 | +({ \ | |
| 246 | + set_thread_flag(TIF_NOTIFY_RESUME); \ | |
| 247 | + false; \ | |
| 248 | +}) | |
| 249 | + | |
| 234 | 250 | struct user_desc; |
| 235 | 251 | extern int do_get_thread_area(struct task_struct *p, int idx, |
| 236 | 252 | struct user_desc __user *info); |
| @@ -334,6 +334,9 @@ static inline void user_single_step_siginfo(struct task_struct *tsk, · 333 unmodified lines | ||
| 334 | 334 | * calling arch_ptrace_stop() when it would be superfluous. For example, |
| 335 | 335 | * if the thread has not been back to user mode since the last stop, the |
| 336 | 336 | * thread state might indicate that nothing needs to be done. |
| 337 | + * | |
| 338 | + * This is guaranteed to be invoked once before a task stops for ptrace and | |
| 339 | + * may include arch-specific operations necessary prior to a ptrace stop. | |
| 337 | 340 | */ |
| 338 | 341 | #define arch_ptrace_stop_needed(code, info) (0) |
| 339 | 342 | #endif |